This tutorial comes from Carson Sievert’s Plotly for R Master Class.

1.1 A case study of housing sales in Texas

The plotly package depends on ggplot2 which bundles a data set on monthly housing sales in Texan cities acquired from the TAMU real estate center. After the loading the package, the data is “lazily loaded”" into your session, so you may reference it by name:

## # A tibble: 8,602 x 9
##    city     year month sales   volume median listings inventory  date
##    <chr>   <int> <int> <dbl>    <dbl>  <dbl>    <dbl>     <dbl> <dbl>
##  1 Abilene  2000     1    72  5380000  71400      701       6.3 2000 
##  2 Abilene  2000     2    98  6505000  58700      746       6.6 2000.
##  3 Abilene  2000     3   130  9285000  58100      784       6.8 2000.
##  4 Abilene  2000     4    98  9730000  68600      785       6.9 2000.
##  5 Abilene  2000     5   141 10590000  67300      794       6.8 2000.
##  6 Abilene  2000     6   156 13910000  66900      780       6.6 2000.
##  7 Abilene  2000     7   152 12635000  73500      742       6.2 2000.
##  8 Abilene  2000     8   131 10710000  75000      765       6.4 2001.
##  9 Abilene  2000     9   104  7615000  64500      771       6.5 2001.
## 10 Abilene  2000    10   101  7040000  59300      764       6.6 2001.
## # … with 8,592 more rows

Let’s see if there’s any pattern in house price behavior over time:

## [1] "gg"     "ggplot"

It’s be nice if we could see which city each line corresponds to when we hover. plotly makes this easy! Just wrap your ggplot object in the ggplotly() function:

If we just want the city name, we can specify exactly what to put in the tooltip:

We can also build plotly objects directly using the plot_ly() function along with dplyr-like syntax. Why would we want to? Well, for one thing, plot_ly() recognizes and preserves groupings created with dplyr’s group_by() function:

## # A tibble: 8,602 x 9
##    city     year month sales   volume median listings inventory  date
##  * <chr>   <int> <int> <dbl>    <dbl>  <dbl>    <dbl>     <dbl> <dbl>
##  1 Abilene  2000     1    72  5380000  71400      701       6.3 2000 
##  2 Abilene  2000     2    98  6505000  58700      746       6.6 2000.
##  3 Abilene  2000     3   130  9285000  58100      784       6.8 2000.
##  4 Abilene  2000     4    98  9730000  68600      785       6.9 2000.
##  5 Abilene  2000     5   141 10590000  67300      794       6.8 2000.
##  6 Abilene  2000     6   156 13910000  66900      780       6.6 2000.
##  7 Abilene  2000     7   152 12635000  73500      742       6.2 2000.
##  8 Abilene  2000     8   131 10710000  75000      765       6.4 2001.
##  9 Abilene  2000     9   104  7615000  64500      771       6.5 2001.
## 10 Abilene  2000    10   101  7040000  59300      764       6.6 2001.
## # … with 8,592 more rows

Since we didn’t specify any mapping, the plot defaults to a scatterplot:

## No trace type specified:
##   Based on info supplied, a 'scatter' trace seems appropriate.
##   Read more about this trace type -> https://plot.ly/r/reference/#scatter
## No scatter mode specifed:
##   Setting the mode to markers
##   Read more about this attribute -> https://plot.ly/r/reference/#scatter-mode
## Warning: Ignoring 616 observations

Let’s change that to a line chart. Similar to geom_line() in ggplot2, the add_lines() function connects (a group of) x/y pairs with lines in the order of their x values and returns the transformed plotly object:

Want to highlight a particular line? Filtering works, and since each add_lines() call returns a pointer to the modified plotly object, we can chain calls together with pipes:

Want to zoom in without losing context? Try a rangeslider():

And just so you don’t think we’re limited to line charts:

## Warning: Removed 616 rows containing non-finite values (stat_bin2d).

Check out The Plotly Cookbook for more details on specific plotly visualization types (“traces”).

Combining views

With htmlwidgets

Since plotly objects inherit properties from an htmlwidget object, any method that works for arranging htmlwidgets also works for plotly objects. In some sense, an htmlwidget object is just a collection of HTML tags, and the htmltools package provides some useful functions for working with HTML tags RStudio and Inc. 2016. The tagList() function gathers multiple HTML tags into a tag list, and when printing a tag list inside of a knitr/rmarkdown document Xie 2016; Allaire et al. 2016, it knows to render as HTML:

This renders two plots, each in its own row spanning the width of the page, because each htmlwidget object is an HTML <div> tag. More often than not, it is desirable to arrange multiple plots in a given row, and there are a few ways to do that. A very flexible approach is to wrap all of your plots in a flexbox (i.e., an HTML <div> with display: flex Cascading Style Sheets (CSS) property). The tags$div() function from htmltools provides a way to wrap a <div> around both tag lists and htmlwidget objects, and set attributes, such as style.

### With shiny Another way to arrange multiple htmlwidget objects on a single page is to leverage the fluidPage(), fluidRow(), and column() functions from the shiny package:

With subplot()

We could also use plotly’s built-in subplot() function to generate a single plotly object with a common y-axis

## Warning: Can only have one: config

Linking multiple views with crosstalk

crosstalk is the R implementation of the powerful crossfilter JS library. Though this dataset isn’t very large, we’ll use it to create a SharedData object that allows us to propagate interaction.

As far as ggplotly() and plot_ly() are concerned, SharedData object(s) act just like a data frame, but with a special key attribute attached to graphical elements. Since both interfaces are based on the layered grammar of graphics, key attributes can be attached at the layer level, and those attributes can also be shared across multiple views. Let’s leverage both of these features to link multiple views of median house sales in various Texan cities:

## Setting the `off` event (i.e., 'plotly_doubleclick') to match the `on` event (i.e., 'plotly_click'). You can change this default via the `highlight()` function.

We can also link different views, like those from ggpairs. Let’s look at the iris dataset because it’s easy on the eyes:

## Warning: Can only have one: highlight

## Warning: Can only have one: highlight

## Warning: Can only have one: highlight
## Setting the `off` event (i.e., 'plotly_deselect') to match the `on` event (i.e., 'plotly_selected'). You can change this default via the `highlight()` function.

Really, we can link any collection of htmlwidgets:

## Adding more colors to the selection color palette.
## Assuming "long" and "lat" are longitude and latitude, respectively
## Setting the `off` event (i.e., 'plotly_deselect') to match the `on` event (i.e., 'plotly_selected'). You can change this default via the `highlight()` function.

We’ll see more about leaflet maps this afternoon.

Your turn!

Try some of these ideas out on your own dataset and see what you can come up with!